CN107045122A - A kind of object detection system and its detection method - Google Patents

Abstract

The invention provides a target detection system and a target detection method thereof. The target detection system includes a radar front-end module, an intermediate frequency signal processing module connected to the output end of the radar front-end module; a filter amplifier circuit connected to the output end of the intermediate frequency signal processing module, An A/D conversion circuit connected to the output end of the filter amplifier circuit; a back-end signal processing module connected to the output end of the A/D conversion circuit; a frequency modulation signal generation circuit connected to the output end of the back-end signal processing module. The target detection method firstly selects the maximum value reference unit; then estimates the background noise power, then calculates the power detection threshold, and finally realizes the target discrimination. The invention solves the technical problems that the false alarm probability and missed detection rate of the existing target detection method are relatively high, and the elimination threshold depends on prior knowledge when obtaining the detection threshold. By eliminating the threshold, the maximum value reference unit can be effectively eliminated, which can effectively improve the detection rate and reduce the missed detection rate.

Description

A target detection system and detection method thereof

technical field

The invention belongs to the technical field of radar target detection, and in particular relates to a target detection system and a detection method thereof.

Background technique

The technical background related to the present invention will be described below, but these descriptions do not necessarily constitute the prior art of the present invention.

With the rapid development of Advanced Driver Assistance Systems (ADAS) and intelligent driving technology, environmental perception and target detection technologies will develop rapidly. The mainstream detection technologies include millimeter-wave radar technology, laser radar technology and vision technology, and impulse radar. Millimeter-wave radar technology is widely used due to its all-weather and low-cost advantages; impulse radar is a type of ultra-wideband radar, and its essence is to use ultra-narrow pulses to achieve target detection.

Target detection method is the core technology of radar detection technology; constant false alarm detection is a commonly used detection method in modern radar systems, constant false alarm detection can automatically adjust the threshold according to the intensity of interference and noise under a certain false alarm probability , which improves the stability and accuracy of radar detection and is widely used in contemporary radar systems. Among them, the false alarm probability is because noise always exists objectively. When the amplitude of the noise signal exceeds the detection threshold, the radar (or other detection system) will be mistaken for the target. This error is called "false alarm". The probability of occurrence is called the false alarm probability.

At present, the target detection methods are mainly divided into two categories: one is the detection method based on the average noise power estimation; the other is the detection method based on the reference unit statistics. Among them, the detection methods based on the average noise power estimation include the Cell Averaging-Constant False Alarm Rate (CA-CFAR) target detection method, the Smallest Of-Constant False Alarm Rate (SO-CFAR) target detection method, and the minimum selection constant false alarm rate (SO-CFAR) method. ) target detection method and the maximum selection constant false alarm (Greatest Of-Constant False Alarm Rate, GO-CFAR) target detection method.

The unit average constant false alarm target detection method takes the average value of all reference unit powers in the reference window as the background noise power estimation value, and estimates the power detection threshold. In the uniform background noise environment, it has better detection performance, but when the reference window length increases When , the test unit may be submerged in the clutter interference and multi-target interference environment, and the unit average constant false alarm target detection method has a high false alarm probability. The minimum selection CFAR and maximum selection CFAR target detection methods are improved on the basis of the unit average CFAR target detection method, and the maximum selection CFAR target detection method is too small for clutter interference and multi-interference targets , the missed detection rate is too high. While the minimum selection constant false alarm target detection method has a high detection rate, but in the environment of clutter interference and multiple interference targets, the false alarm probability is too high.

Target detection methods based on reference cell statistics include Automatic Censored Cell Averaging-Constant False Alarm Rate (ACCA-CFAR) target detection method and Automatic Dual Censored Cell Averaging-Constant False Alarm (Automatic Dual Censored Cell Averaging-Constant FalseAlarm Rate, ADCCA-CFAR) target detection method.

The automatic deletion unit average constant false alarm detection method and the automatic double elimination unit average constant false alarm detection method obtain the detection threshold by eliminating the maximum and minimum reference units in the reference window, and then take the average method, compared with the unit average constant false alarm The detection method has good detection performance, but the determination of the rejection threshold depends on prior knowledge and has limitations.

Contents of the invention

In order to solve the technical problems that the false alarm probability and missed detection rate of the existing target detection methods are too high and when the detection threshold is obtained, the elimination threshold depends on prior knowledge. The invention provides a target detection system and a detection method thereof.

A target detection system, the target detection system includes a radar front-end module, an intermediate frequency signal processing module connected to the output end of the radar front-end module; a filter amplifier circuit connected to the output end of the intermediate frequency signal processing module, and the output of the filter amplifier circuit The A/D conversion circuit connected to the end; the back-end signal processing module connected with the output end of the A/D conversion circuit; the frequency modulation signal generation circuit connected with the output end of the back-end signal processing module; the output of the frequency modulation signal generation circuit The terminal is connected with the radar front-end module; the radar front-end module is used for transmitting and receiving radio frequency signals.

In a preferred embodiment of the present invention, the above-mentioned back-end signal processing module is a DSP/FPGA back-end signal processing module.

In a preferred embodiment of the present invention, the above-mentioned DSP/FPGA back-end signal processing module includes a JTAG debugging unit, an OSC clock unit, an SRAM memory and a FLASH storage circuit. (Please add the block diagram of the module and the connection relationship between each unit)

In a preferred embodiment according to the present invention, a DAC digital-to-analog converter is arranged in the above-mentioned FM signal generating circuit.

In a preferred embodiment according to the present invention, the above-mentioned intermediate frequency signal processing circuit returns the _I1 and I2 _two -way intermediate frequency analog signals returned by the radar front end, after being processed by the filter amplifier circuit, the intermediate frequency is generated by the A/D conversion circuit The digital signal is transmitted to the DSP/FPGA back-end signal processing module and buffered in the SRAM memory at the same time.

In a preferred embodiment according to the present invention, the above-mentioned DSP/FPGA back-end signal processing module configures the DAC digital-to-analog converter through the communication interface to generate a sawtooth wave/triangular wave modulation signal, and process it through an analog filter amplifier circuit to obtain High linearity analog FM signal is provided to the radar front-end to generate radio frequency signal.

The present invention provides a kind of target detection method, described comprising the following steps:

Step 1, select the maximum value reference unit;

Step 2, estimate the background noise power;

Step 3, calculating the power detection threshold;

Step 4, identify the target.

In a preferred embodiment according to the present invention, the maximum value reference unit in the above step 1 is expressed as: x _max =max( _xi ), i∈[1,N]; except the maximum value reference unit in the reference window , the reference unit set of N-1 reference units composed of the rest of the remaining reference units can be expressed as: x _1' , x _2' , x _3' ,...x _N-1 ; then each reference unit is related to β·x _max for comparison.

In a preferred embodiment of the present invention, each of the above-mentioned reference units is compared with β·x _max and the result is set as follows: when all reference units less than or equal to β·x _max form a reference unit set s ₀ ; greater than All reference units of β·x _max form the reference unit set s ₁ , and the elimination of the maximum value reference unit can be expressed by the following formula:

Where k∈[1,N-1], n ₀ represents the number of reference units in the reference unit set s ₀ , 0≤n ₀ <N-1.

In a preferred embodiment of the present invention, the background noise power estimation value in the above step 2 is assumed to be Z, and the calculation formula of Z is expressed as:

Where N _t is an integer, which refers to the size of the reference window minus the maximum acceptable parameter of this target detection method.

The difference between the maximum number of interference target units in the test window, 0<N _{t ≤} N, _xi is the reference in the reference window

unit; x _j is the reference unit in the reference unit set s ₀ .

In a preferred embodiment of the present invention, the power detection threshold in step 3 above is set to T, and the calculation formula of T is: T=Z·α, where α are respectively threshold coefficients.

In a preferred embodiment according to the present invention, in the process of discriminating the target in the above step 4, the test unit is set as x ₀ , and the test unit x ₀ can be judged whether x ₀ is valid by comparing with the power detection threshold T The echo signal unit of the target judges whether the target exists.

In a preferred embodiment according to the present invention, the method for judging whether the above-mentioned target exists is as follows:

In the above formula, H ₀ means no target, and H ₁ means there is a target.

In a preferred embodiment of the present invention, the reference window size N is 24, P _fa is the target false alarm rate; the P _fa =10 ^-4 , the calculation formula of the threshold coefficient α is: α=(P _fa ) ^1/N -1; the parameter β is determined by simulation; the assumption P _D represents the target detection rate.

The advantages of the present invention are as follows:

1. The target detection method provided by the present invention introduces the largest reference unit and obtains the optimal proportional coefficient through a large number of collected actual road condition echo signals through simulation, and then obtains the elimination threshold by multiplying the maximum reference unit power by the proportional coefficient. The coefficient has good applicability; the maximum value reference unit can be effectively eliminated by eliminating the threshold, which can effectively improve the detection rate and reduce the missed detection rate. It overcomes the problem of severe drop in detection rate in the environment of clutter interference and multi-interference targets.

2. The target detection method provided by the present invention selects the corresponding reference unit set according to the number of remaining reference units after removing the maximum value reference unit in the reference window, estimates the background noise power, and then multiplies it by the calculated constant false alarm rate of the target The proportional coefficient is used to obtain the final power detection threshold, which avoids the problem of excessive false alarm probability caused by underestimation of noise power.

Description of drawings

The features and advantages of the present invention will become more comprehensible through the following detailed description provided with reference to the accompanying drawings, in which:

Fig. 1 is the junction principle block diagram of the millimeter-wave radar platform based on in the present invention;

Fig. 2 is the functional block diagram of the processing method of intermediate frequency digital signal in the present invention;

Fig. 3 is the schematic diagram of this target detection method in the present invention;

Fig. 4 is the simulation graph of the detection rate of proportional coefficient and this target detection method among the present invention;

Fig. 5 is the target detection rate simulation curve figure of this target detection method in the uniform noise simulation environment in the present invention;

FIG. 6 is a simulation graph of the target detection rate of the target detection method in the present invention under the non-uniform noise simulation environment.

detailed description

Exemplary embodiments of the present invention will be described in detail below with reference to the accompanying drawings. The description of the exemplary embodiments is for the purpose of illustration only, and in no way limits the invention and its application or usage.

Referring to shown in Fig. 1, the target detection system provided by the present invention comprises radar front-end module, DAC FM signal generation circuit, intermediate frequency signal processing circuit, A/D analog-to-digital conversion circuit and DSP/FPGA back-end signal processing system, wherein DSP/FPGA The signal processing system includes JTAG debugging, OSC clock, SRAM and FLASH storage circuits. The DSP/FPGA signal processing system configures the DAC digital-to-analog converter through the communication interface to generate a sawtooth wave/triangular wave modulation signal, which is processed by an analog filter amplifier circuit to obtain a high-linearity analog FM signal, which is provided to the radar front-end to generate a radio frequency signal . The intermediate frequency signal processing circuit processes the two intermediate frequency analog signals I ₁ and I ₂ returned by the radar front end through a low-noise operational amplifier, and then undergoes A/D analog-to-digital conversion to generate intermediate frequency digital signals, and transmits them to DSP/FPGA for signal processing In the system, the buffer is in the SRAM memory. The radar front-end module (including microstrip antenna) is used to transmit and receive radio frequency signals.

Referring to Fig. 2, the processing method of the intermediate frequency digital signal is given. The intermediate frequency digital signal buffered in the SRAM memory, after A/D conversion, is temporarily stored in the first-in-first-out buffer (First Input First Output, FIFO), and then digitally filtered and Fast Fourier Transform (Fast Fourier Transform) is performed on it. Transformation, FFT), sent to the detector for the target detection method to extract.

Referring to Figure 3, this figure shows a target detection method. Among them, I ₁ and I ₂ are radar echo signals, and the detection envelope sequence obtained after passing through the square law detector is used as the input signal of this detection algorithm. x _max is the maximum value reference unit selected from the reference window; β is the proportional coefficient, 0<β≤1; x ₀ is the test unit, x ₁ , x ₂ , x ₃ ,... x _N is the input from the envelope sequence The reference window reference unit sequence extracted from the signal, N is the size of the reference window. The execution steps of the detection method are as follows:

In step 1, the maximum value is selected as a reference unit, and x _max can be expressed as:

x _max = max(x _i ), i∈[1,N] (1)

Step 2 Background noise power estimation. In addition to the maximum value reference unit in the reference window, the reference unit set of N-1 reference units composed of the remaining reference units can be expressed as: x _1' , x _2' , x _3' ,...x _N-1 , and then Each reference cell is compared to β·x _max separately. All reference units less than or equal to β·x _max form a reference unit set s ₀ , and all reference units greater than β·x _max form a reference unit set s ₁ , the elimination of the maximum value reference unit can be expressed by the following formula:

Where k'∈[1,N-1], n ₀ represents the number of reference units in the reference unit set s ₀ , 0≤n ₀ <N-1.

Assuming that Z is the estimated value of the background noise power value, the calculation formula of Z is expressed as:

Where N _t is an integer, which refers to the difference between the size of the reference window minus the maximum number of interference target units in the reference window that this target detection method can bear, 0<N _{t ≤} N, and x _i is the reference unit in the reference window ; x _j is the reference unit in the reference unit set s ₀ .

According to the comparison of n ₀ and N _t value, the detection method will select the corresponding reference unit from s ₀ or the reference window for estimating the background noise power Z. When n ₀ ≥ N _t , the interference target reference unit has a higher probability of appearing in the reference unit set s ₀ , in order to improve the detection rate of the target detection method, the reference unit will be selected from the reference unit set s ₀ for estimating the Z value . When n ₀ <N _t , the probability of the interference target reference unit appearing in the unit set s ₀ is small. In order to avoid the Z value estimation being too small, resulting in a high false alarm rate, all the reference units in the reference window are selected for estimation Background noise power Z.

Step 3 Calculation of the power detection threshold, the background noise power value Z is multiplied by the threshold coefficient α to obtain the power detection threshold T, and the calculation formula of T is:

T=Z·α (4)

Among them, α are the threshold coefficients respectively.

Step 4 Target discrimination. The test unit x ₀ can judge whether x ₀ is the echo signal unit of a valid target by comparing it with the power detection threshold T, that is, it can judge whether the target exists. Suppose H ₀ means no target, H ₁ means there is a target, and the method of judging whether the target exists as follows:

In this method, the reference window size N is taken as 24, and P _fa is the target false alarm rate. In this method, P _fa =10 ^-4 , and the calculation formula of the threshold coefficient α is:

α=(P _fa ) ^1/N -1 (6)

The parameter β can be determined by simulation, assuming that _PD represents the target detection rate.

Refer to Fig. 4, which is a simulation curve corresponding to the proportionality coefficient β and _PD . It can be seen from the simulation curve that when β=0.04, the detection method has a high detection rate under both uniform and non-uniform background noise. In the uniform noise environment, P _d reaches a maximum of 95.13% when γ = 0.01; when the value of γ increases from 0 to 0.042, P _d decreases from 95.12% to 94.50%. In the environment of non-uniform background noise, when the value of β increases from 0 to 0.042, P _d increases from 43.30% to 79.46%; when β = 0.042, P _d reaches the maximum of 83.68%; P _d is reduced from 78.95% to 22.11%. Considering the detection performance of this detection method in uniform and non-uniform noise environment, γ is taken as 0.042. According to the parameters α and β determined by calculation and simulation, the detection performance of this detection method and other detection methods were simulated and analyzed in the environment of uniform and non-uniform background noise by Monte Carlo method in Matlab environment.

Referring to FIG. 5 , this figure is a simulation comparison curve of the detection rate of the detection method of the present invention, the ADCCA-CFAR target detection method and the ACCA-CFAR target detection method in a uniform background noise environment. It can be seen from the figure that the detection rate of this target detection method reaches 97.45%, which is close to the target detection methods of CA-CFAR, ADCCA-CFAR and ACCA-CFAR. When the detection probability is 80.00%, compared with the CA-CFAR method, the loss of this target detection method is about 0.06dB, and the loss of ACCA-CFAR and ADCCA-CFAR method is about 0.12dB, which shows that this detection method has good performance in the uniform noise environment. Better detection performance.

Refer to Fig. 6, which is a simulation comparison curve of the detection rate between this target detection method and the ACCA-CFAR and ADCCA-CFAR target detection methods in a non-uniform background noise environment. It can be seen from the figure that the detection rate of this target detection method is better than that of ACCA-CFAR and ADCCA-CFAR detection methods. When SNR=25dB, the detection rate of this target detection method is 96.71%, which is higher than ACCA-CFAR and ADCCA respectively. - The detection rate of CFAR method is 10.52% and 5.12%, which shows that the target detection method also has good detection performance in the non-uniform noise environment.

Although the present invention has been described with reference to exemplary embodiments, it should be understood that the present invention is not limited to the specific embodiments described and shown in detail herein, and that it is possible for those skilled in the art to do so without departing from the scope defined by the claims. Personnel may make various improvements or modifications to the exemplary embodiments.

Claims (14)

1. a kind of object detection system, it is characterised in that：The object detection system includes radar front end module, with radar front end The IF signal processing module of the output end connection of module；The filter and amplification electricity being connected with the output end of IF signal processing module Road, the A/D change-over circuits being connected with the output end of filter amplification circuit；The rear end letter being connected with the output end of A/D change-over circuits Number processing module；Circuit occurs for the FM signal being connected with the output end of back end signal processing module；The FM signal occurs The output end of circuit is connected with radar front end module；The radar front end module is used to launch and receive radiofrequency signal.

2. object detection system according to claim 1, it is characterised in that：The back end signal processing module is DSP/ FPGA back end signal processing modules.

3. object detection system according to claim 2, it is characterised in that：The DSP/FPGA back end signals processing module Including JTAG debugging units, OSC clock units, SRAM memory and FLASH storage circuits.

4. object detection system according to claim 2, it is characterised in that：The FM signal occurs to be provided with circuit DAC digital analog converters.

5. object detection system according to claim 3, it is characterised in that：Before the signal processing circuit of intermediate frequency is to radar Hold the I returned₁And I₂Two-way analog intermediate frequency signal, after amplifying circuit is handled after filtering, then by the production of A/D change-over circuits Raw digital intermediate frequency signal, and be transferred in DSP/FPGA back end signal processing modules, simultaneous buffering is in SRAM memory.

6. object detection system according to claim 4, it is characterised in that：The DSP/FPGA back end signals processing module DAC digital analog converters are configured by communication interface, sawtooth waveforms/triangular modulation signal are produced, and pass through analog filtering Amplifying circuit processing, obtaining the analog fm signal of high linearity, there is provided produce radiofrequency signal to radar front end.

7. a kind of object detection method, it is characterised in that：Comprise the following steps：

1) maximum reference unit is chosen；

2) Background Noise Power is estimated；

3) power detection thresholding is calculated；

4) target is differentiated.

8. a kind of object detection method according to claim 7, it is characterised in that：Maximum reference in the step 1 Unit is expressed as：x_max=max (x_i),i∈[1,N]；In reference window in addition to maximum reference unit, remaining residue is with reference to single The reference unit collection of N-1 reference unit of member composition is represented by：x_1',x_2',x_3',……x_N-1；Then each reference unit Respectively with β x_maxIt is compared.

9. a kind of object detection method according to claim 8, it is characterised in that：Each reference unit respectively with β·x_maxIt is compared result setting as follows：When less than or equal to β x_maxAll reference units composition reference unit integrate as s₀； More than β x_maxAll reference units composition reference unit integrate as s₁, the rejecting of maximum reference unit can use below equation It is expressed as follows：

Wherein k' ∈ [1, N-1], n₀Represent reference unit collection s₀In number of reference, 0≤n₀＜ N-1.

10. a kind of object detection method according to claim 9, it is characterised in that：Background Noise Power in the step 2 Estimate is assumed to be Z, and the calculation formula of the Z is expressed as：

Wherein N_tFor integer, refer to reference windows size subtract it is maximum in the reference window that this object detection method can at most be born The difference of jamming target unit number, 0 ＜ N_t≤ N, x_iFor the reference unit in reference window；x_jFor reference unit collection s₀In ginseng Examine unit.

11. a kind of object detection method according to claim 10, it is characterised in that：Power detection door in the step 3 Limit is set to T, and T calculation formula is：T=Z α, wherein α are respectively threshold coefficient.

12. a kind of object detection method according to claim 11, it is characterised in that：Target mistake is differentiated in the step 4 Cheng Zhong, sets test cell as x₀, the test cell x₀It can differentiate x by being compared with power detection thresholding T₀Whether it is to have The echo-signal unit for imitating target is that judgement is that target whether there is.

13. a kind of object detection method according to claim 11, it is characterised in that：The target whether there is determination methods such as Under：

H in above-mentioned formula₀Indicate no target, H₁Indicate target.

14. a kind of object detection method according to claim 11, it is characterised in that：The reference window size N takes 24, P_fa For target false alarm rate；The P_fa=10^-4, threshold coefficient α calculation formula is：α=(P_fa)^1/N-1；The parameter beta passes through emulation It is determined that；The hypothesis P_DRepresent target detection rate.